Abstract Substance use disorders have a large degree of co-morbidity with psychiatric disorders including bipolar disorder, schizopherina, and depression. A key challenge to understanding the genetic basis of substance use disorders is to understand at a genetic level its relationship with psychiatric disorders. Whole genome sequencing data of individuals with substance use and psychiatric disorders has the potential to provide extensive information on the relationship between them. However effectively interpreting variants from such data particularly in the vast non-coding regions of the human genome will require novel computational approaches to better annotate the human genome. We will develop several approaches to produce a more relevant annotation of the human genome for interpreting such whole genome sequencing data. One limitation of existing epigenome based annotations of the genome for relevant samples from brain regions is they are derived from a complex mixture of cell types. We will define epigenome annotations such as chromatin states computationally at a single cell type level by deconvoluting population based ChIP-seq data in a framework work that integrates single cell RNA-seq data and Hi-C or other information associating distal regions with genes. We will also develop approaches to better map highly relevant epigenomic data on substance use disorders from model organisms to human through a novel approach that learns a mapping based on common activity from a compendium of existing epigenomic data. We will also develop approaches that will learn from high-throughput functional testing genomewide predictions of the functionally important positions and variants even in cell types not tested using epigenomic features and specially constructed sequence features that will generalize across cell types. Through collaborations the genome annotations produced here will be applied to analyze multiple whole genome sequencing data sets of individuals with substance use disorders, psychiatric disorders, or both. We will identify annotation classes as being associated specifically with variants of substance use disorders, psychiatric disorders, or jointly between them to gain biological insights into the biological relationship between the disorders. All computational methods developed and genome annotations produced will be broadly disseminated.